Our concern is to explain how and why models give us useful knowledge. We argue that if we are to understand how models function in the actual scientific practice the representational approach to models proves either misleading or too minimal—depending on how representation is defined. By ‘representational approach’ we mean one that attributes the epistemic value of models to the representational relationship between a model and some real target system. In contrast we propose turning from the representational approach to the artefactual, which also implies a new unit of analysis: the activity of modelling. Modelling, we suggest, could fruitfully be approached as a scientific practice in which concrete artefacts, i.e., models, are constructed with specific representational means and used in various ways, for example, for the purposes of scientific reasoning, theory construction and design of experiments and other artefacts. Furthermore, we propose that in the activity of modelling the construction of models is intertwined with the construction of new phenomena, concepts, and theoretical principles. We will illustrate these claims by studying the construction of the ideal heat engine by Sadi Carnot.
- Modelling – Engineering sciences – Representation – Carnot – Epistemic tools – Heat engine